(ag
THE GROWTH OF METAL WHISKERS
‘ON ELECTRICAL COMPONENTS:
8. M. Arnold
Bell Telephone Laboratories, Inc.
Murray Hill, New Jersey
Introduetion
Bell Telephone Laboratories has had an
active program devoted to studies of the
nature of metal whisker growth since a
number of tranemission troubles were
traced directly to the presence of conduct-
ing fllaments on the surfaces of some close:
ly spaced parte in telephone equipment about
ten yeare ago. Prior to actual failure,
characterized by low impedance to ground
in equipment which had been operating
satisfactorily for many months, there had
been no indication of equipment degradation.
Failure of the hermetically sealed units
evidently had occurred because of some-
thing not present at the time of installation
and was not a result of a progressive com-
ponent deterioration such as might be a:
Sociated with corrosion or absorption of
moisture, Identification of the part the
{laments played in the equipment failures
and the subsequent findings that the fila~
ments were metallic and actually required
a period of time to develop, indicated that
troubles of this kind would not be confined
toa particular type of apparatus. Such
failures might be expected at sonie in-
definite time in any equipment in which
specific metals, small spacings, low volt~
ages and critical circuits were involved.
‘These indications have been confirmed,
and numerous failures since have been
traced to the presence of metallic whisker
Where high voltages exist destruction of
any bridging whiskers automatically occurs
but their importance must certainly in-
crease ae low voltage circuitry and micro-
scopic spacings become adopted ever more
widely.
This report has been written to illustrate
the variety of equipment types in which
metal whiskers have been found, to describe
number of the actual service failures
traced to the presence of these whiskers
and to suggest possible methods of prevent
ing such failures,
Components Associated with Whisker Growth
Of interest because it was the first part
of the telephone plant in which growths of
conducting filaments are known to have
caused electrical failure, is the channel fre
quency filter shown in Fig. 1. In this unit,
essential for the simultaneous transmission
of a number of conversations over a multi-
channel telephone line, whiskers developed
on zine-plated brackets and bridged across
to nearby air capacitors, as shown in Fig, 2.
Approximately one year was required from
Fig. 1. Channel frequency filter, Pencit indiestes
bridged by whiskers.
Fig. 2. Zine whiskers responsible for low impedance in
channel filter. Spacing beidged approximately 3/16 in.1%
the time of astembly until a low impedance
condition developed as a result of whiskezs
spanning this 3/16 in. spacing.
Shortly after the troubles in the channel
filters had been traced to the presence of
such filaments, a failure of a sensitive
potentiometer also was identified with
metal whisker growth. Here the whiskers
had not developed on the potentiometer it-
self, however, but on the inside of the
potentiometer enclosure. The enclosure
Was made of tin-plated steel, and whiskers
had grown on the internal tin coating,
Some of the whiskers had become dislodged
and fallen acroes several turns of the
potentiometer winding.
What may have been a somewhat similar
situation arose with regard to a number of
carbon resistors, This type of resistor is
made by depositing a film of pyrolytically
formed carbon onto a ceramic core. The
completed resistor in effect consists of
core coated with a spiral ribbon of carbon,
Wath adjacent turns separated from each
other by a narrow spiral path free of
carbon. Positive contact is obtained at
the ends of the core by means of silver
paste and brass ferrules, The ferrules
themselves are tin plated to insure good
contact with the mounting clips, Whiskers
were found on a number of the ferrules.
Nonpereiatent troubles had been experienced
with some of the resistors, and it was
believed that whiskers from the ferrules
had become detached and lodged across the
gap separating adjacent turns of the carbon
fm,
A copper oxide rectifier destined for
military uae failed in preservice triale
because of the growth of metal whisker;
Figure 3 de an enlarged view of a portion
of the rectifier.* The whiskers can be
seen bridging between the tin coated texmi-
pale, short cizcuiting the unit. Because it
had been enclosed, the actual trouble was
found only after the unit had been di:
mantled, an operation not slways possible
or desirable under field conditions
In Fig. 4 is shown a part of a pro-
tector mounting which failed because of
whisker growth. In the assembly of this
particular mounting, a clamping screw
passes through a cylindrical bushing and
forces the end of the bushing into close
contact with a plated terminal lug.
‘This te an example of a situation in which
the likelihood of trouble due to whiskers is
enhanced by assembly requirements. It ie
believed that whisker growth is a result of
stress, Further, it has been shown that
pyresure accelerates the groweh of whiskers,
Tn the pileup in question, compression
stresses are exerted by the clamping screws
and a high preasure is developed at the edge
of the bushing where it bears against the
terminal face, There is sufficient space at
the side of the bushing for whiskers develop-
ing in the highly stressed area to pass
through to the adjacent terminal and estab-
Lsh a low impedance path, In Fig. 4 can
be seen a few of the relatively long whiskers
which developed at the edge of the high
pressure area immediately adjacent to the
rim of the bushing. Also, there are present
a few of the shorter whiskers characteristic
of those observed on other parts of the
surface outside of the pressure area.
‘The terminal strip is a frequently used
component in a wide variety of equipment
[rip available in many forms, but often
consists essentially of a series of closely
spaced metal fingers or lugs protruding
from a strip of insulating material, A
hot-dipped or electroplated coating of tin
or tin-lead is usually applied to insure ease
in making soldered joints. In Fig. 5 an
enlarged view of & portion of a unit with tins
plated terminals makes clear the reason
occasional failures are encountered.
Fig. 3. Tin whiskers growth in toubl
‘oxide rectifier (3).
‘+ The original negtfications indicated in the figure captions have been subjected to vatious reductions in
raking the halfioe illustrations
* The BeiterFig. d, Protector sounting tera
pressure area (1580.
Fig. §. Tin whiskers on individual menbers of terminal
strip 20.
‘There have been some instances where
metal whiskers developed on relay mem-
bers. The magnetic iron parte require a
surface finish as protection against cor~
rosion, Electroplated nickel or nickel plus
chromium may be ued, but frequently zinc
or cadmium is applied, The contact springs
often are tin-coated at the clamped ends to
facilitate soldering, and the pileup pressure
is applied by means of screws bearing on
compression plates which may be coated
with tin, zine or cadmium, Either zinc or
cadmium may be applied to the armature
7
return spring. In Fig, 6 is shown a relay
contact spring and some of the whiskers
responsible for a grounded circuit, Figure
7 shows how whiskers grew on one of the
compression plates of a relay and came into
contact with an adjacent spring. This situ-
ation ia similar to that described in the
discussion of the short-circuited protector
mounting where stress accelerated whisker
growth.
Another type of failure attributed to
whisker growth took place ina relay oper-
ating in a eireuit handling some 200 volts.
A persistent arc caused considerable
damage to the relay. Examination dis.
closed the presence of numerous whiskers
fon contact supports, It is believed that
whiskers had bridged between closely
spaced parts to initiate an arc and, once
initiated, the arc had been maintained by
the standing voltage until protection d
vices operated to cut off current flow.
Fig. 6 Whiskers on tisepleted relay contact spring (12X).
A relay armature return spring, electro
plated with cadmium, Je shown in Figs. 8
and 9. An appreciable growth of whiskers,
some of them quite long, has taken place.
Here, of course, they will cause no trouble
as long ae none becomes lodged at some
vulnerable area.78
Fig. 9. Whiskers on cadmium plated armetute ttuen spxing
of relay shown in previous figure (30X).
Another widely used element is the vari-
able air capacitor. This too has at times
been the site of whisker trouble and was re-
ferred to as far back as 1946, (2) while
aluminum frequently is specified in the
Fig. 7, Tin whiskers along edge of high-pressure atea
Spring pileap orea of relay (15%).
Fig, 10. Heavy groath of whisk
Fle. 8 Small rolay (1.5%), vstiable sir capacitor (34).
on end plate ofPig. UL. Whiskor
vious figure (6X).
reciting cepecitor shown in pre
fabrication of these capacitors, both brass
and steel are used and may be left unfinished
or coated with cadmium oF zine, A portion
of one such variable capacitor with @ fuzz
of whiskers on the plated end piece is shown,
in Fig. 10, In Fig, 11 can be seen the
whisker growths which led to failure when
contact was made with an adjacent plate,
Whiskers have also been observed in
metallized capacitor paper. The capaci-
tore are made by winding many turns of
two sheets of the metallized paper to pro-
duce a closely wrapped spiral sandwich of
alternate layers of paper and metal. El-
ectrical contact is obtained at the ends of
the two metal films by means of projecting
tabs soldered to flexible leads or to termi+
nal posts. Samples of capacitor papers
coated with zine or cadmium have been
examined, and on a number of them metal
whiskers have been found,
While there have been no reports of
failures which could be identified with
whisker growth in this type of unit, the
question of the possibility of such failure
has been raised frequently enough to warrant
a brief comment. There is no evidence
that growing whiskers will penetrate the
interleaving paper used in capacitors,
9
Actually, whiskers have been grawn in con
tact with such paper, but they were deflected
without penetration, Only if the metal film
extended to the paper edge so that projecting
whiskers could make contact with the can
cover or leads might there be 2 possible,
although unlikely, trouble condition.
Because the function of a chassis is to
serve purely as a structural member, it
may be overlooked complely when relia~
Dility is under consideration. However,
many chassis are fabricated from steel
electroplated with cadmium or zine which
have a tendency to develop whiskers. Ina
cadmium -plated chassis examined some two
yeare after purchase, whiskers were found
fo exist over the entire surface, Further-
more, this was not an isolated case. Other
chassis have been seen with comparable
growths. The presence of whiskers on a
chassis would be usually of academic
interest but there is the possibility that
whiskere may be detached and find a rest=
ing place in critical cireuit area, Further-
more, should a conducting member of a cir-
cuit inadvertently be located or displaced so
that it is only a short distance from the
chassis, contact could easily be made in an
area of dense whisker growth.
‘Also of interest is the component can
cover used to minimize the deposition of
dust, reduce air flow and gaseous contami-
nation and to serve as electrical or mag-
netic shielding. Although a cover may not
be an integral part of the electrical circuit,
it may be teaponsible for troubles connected
with metal whiskers, Many can covers are
electro-finished with corrosion-protecting
Coatings of zine or cadmium or may be tin=
coated because of solderability require
ments, Growths of whiskers on such sur-
faces are not uncommon and, as pointed out
in the discussion of the failure of @ potentio-
meter, may be responsible for the mal=
functioning of a component if dislodged
whiskers get to the wrong places. Whisker
growthe have been found on the internal
Surfaces of cans used as capacitor and relay
covers, transistor enclosures and small
instrument housings. Cadmium whiskers
in thermostat enclogures have been referred
to in the literature, (3)
Of related interest are several unite that
were enclosed in plated brass or ateel cans
but where a different situation developed.
Here troubles arose because of whisker
growth not in the inside, but on the outside
Surfaces of the cans, either ona lip or a
rolled edge separated only @ short distance
from a terminal or a load, Whiskers grew
long enough to bridge these intervening80
gaps. Asa result, both resistors and
tubular capacitors have required servicing
to correct low impedance conditions.
Failures of a number of selector switches
have been caused in a similar manner, The
switch case normally is electrically sepa-
rated from the contact members by means
of an ingulating disk. However, in some
instances whiskers growing on the case have
been long enough to extend beyond the disk
and make physical contact with an exposed
terminal
Individual unite of many assemblies are
interconnected by means of leads or cables
terminated by connectors such as the AN
type. Frequently these connectors are
made of aliminum alloys but are given a
Gnal tin coating, A number of low-voltage
circuits have been examined after low im-
pedances were reported. It was found that
Whisker growth had developed on the inside
surfaces of the connector shells and that
some of the whiskers had bridged across to
individual contact elements.
Evacuation or maintenance in an atmos~
phere of inert gas will not prevent the
development of whiskers on a component,
In Fig. 12 is shown an evacuated glass.
enclosed quartz crystal resonator. Whiskers
which grew on the crystal tin-plated support
wires are clearly visible in Fig. 13.
Similar growths have been observed in other
Fig. 13. Tin whiskers growth on support wires in resonator
us,
Fig. 14. Whiekote growth on solder fillet ia resonator (48%),
unite maintained under nitrogen at low
pressure. The solder fillet at the attach-
ment area of the crystal in one of these units
is shown in Fig. 14. Whiskers have develop-
ed on the solder to a length sufficient to
reach the edge of the quarts, made contact
with similar whiskers growing from the
fillet on the opposite face, and short circuit
the unit.
Reference has just been made to the oc~
currence of whiskers at the joint in the
quartz crystal resonator. The growth of
whiskers from soldered joints in genezal
has not occurred; it ig believed that in thisinstance the high-frequency vibration of the
piezoelectric quartz produced stresses in
the solder joint and initiated whisker growth.
Laboratory studies of identical units have
shown whisker growth only at the joint of
crystals being vibrated.
Immersion of a component part in oil
also has been suggested as a means of pre-
venting the formation of whiskers. Figure
15, however, shows that a tin-coated bracket
immersed in a medium viscosity hydro-
carbon oil immediately after it was electro-
plated is covered with a substantial growth
of whiskers,
Fig. 1. Whiskers growth from tin-plated part submerged La
ott 22%).
‘The use of a duplex electroplate (first
cadmium and then tin} is etill another
suggested approach to the problem of obtain-
ing finishes not susceptible to whisker
growth. This coating, however, will not
insure the absence of whisker growth,
Whiskers may develop on tin deposited over
cadmium or even from tin codeposited with
cadmium, A number of coatings of this
type are currently under study and will be
discussed ina later paper.
Precautior
-y Measures
There are various procedures that may
be followed to reduce or prevent whisker
growth, although at present no universally
applicable solution is known. Circuit ele~
ments must be considered on an individual
8
basis with respect to the need for preventive
measures, A service interruption due to
whisker growth may be merely an annoy-
ance, to be remedied if and when it occurs:
oF an interruption may be catastrophic, to
be prevented regardless of the effort in-
volved. This last may require the use of
a different metallic finieh or the redesign
of a component for complete assurance of
freedom from whisker troubles,
Because of the susceptibility to whisker
growth of coatings of tin, zinc and cadmium
it may be necessary to curtail their use on
specific components. In many instances,
it will be feasible to'use substitute metallic
finishes or even make use of nonfinished
parte in critical areas. Thus, electro-
doposited nickel, chromium o* copper may
be specified. If ease of solderability is a
requirement, silver or gold may be plated
in selected areas, The substitution of
silver is subject to some restriction, how-
ever, in that silver migration into insulation
is a possibility under the influence of a
standing potential, and silver may develop
a filamentary growth, "hair silver," in an
atmosphere of volatile sulfur compounds,
Phosphor bronze, brass or other copper
alloys, corrosion-resistant steels, aluminum,
or magnesium alloys often may be left un-
finished,
Enhanced whisker growth is to be expect
ed on the thinner electrodeposited coatings.
The initially deposited layers of a plated
coating are known to be highly stressed and,
as mentioned previously, all of our experi-
mental evidence to date points to 2 tre
induced whisker growth mechaniam, The
internal stresses decrease with increase in
coating thickness, and the heavier coatings
therefore are to be preferred when the
possibility of whisker growth must be con-
sidered,
Ifa tin coating be fused after electro-
plating, there will be a marked reduction in
whisker growth, probably because of the
opportunity afforded for stress relief,
‘The use of hot-dipped, rather than
electroplated, coatings also will result in
a major reduction in the growth of whiskers,
Several factors are involved here--the
coatings are usually thicker, there may be
the formation of an alloy layer with the
basie metal, and the metal deposited in the
molten condition is probably less highly
stressed.
As pointed out in the discussions of the
whiskers in the relay spring pileup, com-
pressive stresses greatly accelerate
whisker growth, Thus, whisker growth82,
may be anticipated on a tin-coated surface,
for instance, where it is subjected to a high
bearing pressure from another member,
A barrier may be required between closely
spaced circuit elements in such an area if
high static stresses cannot be avoided,
Where tin-coated parts are involved,
maintenance of low relative humidity and
low temperature will reduce, although not
entirely prevent, the growth of whisker
Conclusion
Obviously there are multitudes of
electrical circuits which are functioning
satisfactorily and which will continue to do
80 without reason to suspect untoward
difficulties. Nevertheless, 20 has been
shown inthis discussion, metal whiskers
have been associated with a rather diverse
group of components. While no attempt
has been made to compile a complete list,
enough exarnples have been given to indicate
the range of component types which have in
the past, and may in the future, provide
opportunities for circuit troubles because
of whisker growth. Particularly in closely
spaced or miniaturized equipment, design
and metal finish cannot be specified with
quite the freedom allowed where widely
spaced parts are involved,
Precautionary measures are possible,
and a number have been suggested. While
the possibility of whisker growth troubles
may seem remote in many installations,
some consideration should be given to the
question of what may happen and how serious
would be the consequences should a spacing
be bridged by a conducting filament.
1. RM, Fisher, L.S. Dazken and K.G.
Garroil, "Accelerated Growth of Tin
Whiskers," Acta Metallurgica, 2 (1954)
368.
2. HL. Cobb, "Cadmium Whiskers," The
Monthly Review, Am. Blectroplaters
Boe 33 Teneo ene
3. J.B. Newkirk, "Metal Whiskers in
Automatic Blanket Thermostats," Metal
Progress, 68 (1955) 88,